Flavor-related biochemical and volatile compound data of sturgeon meat under different thermal processing microenvironment
Description
This dataset contains the original and processed data supporting the above article. It comprises eight main figures, eight supplementary figures, and seven supplementary tables, covering descriptive sensory evaluation, electronic nose and electronic tongue responses, free amino acid composition, IMP content, and equivalent umami concentration (EUC), HS-GC-IMS volatile fingerprints, and the multivariate/machine-learning models linking chemical profiles to sensory attributes. All experiments were performed on frozen-thawed sturgeon meat subjected to three processing methods (steaming, boiling, and sous-vide ) at three temperature levels (50, 70, and 100 degrees C), with a raw untreated control.
Files
Steps to reproduce
Sample preparation and thermal processing. Frozen sturgeon (Acipenser spp.) fillets from a commercial aquaculture farm (Handan, Hebei, China) were stored at -18 °C, thawed at 4 °C, portioned (10.0 g) and assigned to ten groups (n = 3): RAW and steaming, boiling or sous-vide at 50, 70 or 100 °C (S50-S100, B50-B100, V50-V100). Steaming used a stainless-steel steamer (Supor, China), boiling deionized water (1:50, w/v) and sous-vide vacuum-sealed bags (-0.1 MPa, 80 µm) in a water bath. Core temperature was monitored with a thermocouple (TASI TA612C, China); samples were cooled on ice and held at 4 °C (<2 h). Sensory evaluation. Eight trained assessors completed six 2-h sessions (ISO 8586:2012). Eight attributes (fishy, fatty, umami, sweet, bitter, tenderness, juiciness, overall acceptance) were scored on 10-cm unstructured line scales; 1 × 1 × 1 cm samples were served in randomized order (22 ± 1 °C), in triplicate. Instrumental analyses. E-nose (PEN3, Airsense Analytics, Germany; 10 MOS sensors): 2.0 g sample, 20-mL vial, 25 °C, 30 min. E-tongue (TS-5000Z, Insent, Japan; seven lipid/polymer membrane sensors): filtered aqueous extracts (5.0 g/100 mL). Free amino acids (18): 5% (w/v) sulfosalicylic acid extraction; L-8900 analyzer (Hitachi, Japan), ninhydrin derivatization (570/440 nm). IMP: 5% (v/v) perchloric acid extraction (4 °C, 30 min), pH 6.5-6.8; HPLC (1260 Infinity, Agilent, USA), ZORBAX Eclipse XDB-C18 column (4.6 × 250 mm, 5 µm), UV 254 nm, 0.05 M KH2PO4 (pH 4.5)/methanol (95:5), 0.8 mL/min. Volatiles: HS-GC-IMS (FlavourSpec, G.A.S. mbH, Germany); 2.0 g sample, 60 °C, 20 min; 500 µL headspace injected at 85 °C onto an MXT-5 column (15 m × 0.53 mm, 1.0 µm; Restek, USA); drift tube 45 °C, 500 V/cm; identification against GC-IMS library and NIST 2020. Derived indices. EUC (g MSG/100 g) per Yamaguchi (1967) (Asp 0.077, Glu 1.0, IMP 1.0); TAV and OAV as concentration/threshold ratios. Four-tier biomarker screening: OAV >= 1; PLS-DA VIP >= 1; ANOVA P < 0.05; |Spearman r| >= 0.6 with a sensory attribute. Statistics and software. Two-way ANOVA with Tukey HSD (α = 0.05) and dual-level significance lettering (paired t-test, Bonferroni correction). PCA used Z-score data; PLS-DA validated by permutation testing (n = 500); PLSR with leave-one-out cross-validation for sensory prediction; Spearman correlation for networks (|r| > 0.7, P < 0.05); random forest (n_estimators = 100) with SHAP. Computations used VOCal v1.4, LAV v2.3 and Python 3.10.